The role of verifiability and privacy in the strategic provision of performance feedback: Theory and experimental evidence

We theoretically and experimentally analyze the role of verifiability and privacy in strategic performance feedback using a 'one principal-two agent' context with real effort. We confirm the theoretical prediction that information transmission occurs only in verifiable feedbackmechanisms and private-verifiable feedback is the most informative mechanism. Yet, subjects also exhibit some behavior that cannot be explained by our baseline model, such as telling the truth even when this will definitely hurt them, interpreting 'no feedback' more optimistically than they should, and being influenced by feedback given to the other agent. We show that a model with individual-specific lying costs and naive agents can account for some, but not all, of these findings. We conclude that in addition to being naive, some agents also suffer from self-serving biases and engage in non-Bayesian social comparisons in their interpretation of performance feedback.